AI Paper Atlas

Research map / Reinforcement learning / Policy gradient and actor-critic

Counterexample Guided RL Policy Refinement Using Bayesian Optimization

Briti Gangopadhyay, Pallab Dasgupta

NeurIPS 2021 · Reinforcement learning › Policy gradient and actor-critic · approach: Safety layers and shielding · 6 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
refine the PPO policy with counterexample-guided updates instead of training from scratch with a violation penalty (Policy A)PPO · Cart-polefailures: 179 -> 0Improved
refine with counterexample-guided updates instead of Policy APPO · Pendulumfailures: 89 -> 89No clear effect
refine with counterexample-guided updates instead of Policy APPO · BipedalWalkerfailures: 45 -> 0Improved
refine with counterexample-guided updates instead of Policy APPO · LunarLander Continuousfailures: 42 -> 0Improved
refinement instead of Policy B (scratch training on only counterexample traces)PPO · Pendulumfailures: 102 -> 89Improved
refinement instead of Policy BPPO · BipedalWalkerfailures: 145 -> 0Improved
refinement instead of Policy C (scratch with original + counterexample traces)PPO · BipedalWalkerfailures: 41 -> 0Improved
refinement instead of Policy CPPO · LunarLander Continuousfailures: 5 -> 0Improved
refinement instead of Policy CPPO · Cart-poletraining steps: 1M -> 150KImproved
refinement instead of Policy CPPO · BipedalWalkertraining steps: 8M -> 2.8MImproved
refinement instead of Policy APPO · Pendulumtraining steps: 1.6M -> 850KImproved
add penalty term for property violation when training from scratch (Policy A) without exposing parameter variationsPPO · Cart-polefailures: 179Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy